# Security Policy ## Scope This repository contains **documentation and educational templates** for Claude Code. It does not include executable code that processes user input or runs in production environments. **Security concerns specific to this repository:** - Documentation accuracy for security practices - Template code quality and security patterns - Threat database integrity ([`machine-readable/threat-db.yaml`](./machine-readable/threat-db.yaml)) **Out of scope:** - Security vulnerabilities in Claude Code CLI itself → Report to [Anthropic](https://github.com/anthropics/claude-code/security) - Security issues in MCP servers → Report to respective server maintainers ## Reporting a Security Issue If you discover a security concern related to this guide (examples: malicious template, incorrect security advice, threat database inaccuracies), please: 1. **Email**: florian.bruniaux@methode-aristote.fr - Subject: `[SECURITY] Claude Code Guide - Brief Description` - Include: Affected file/section, description, impact assessment 2. **GitHub Private Disclosure**: Use [Security Advisories](../../security/advisories/new) for sensitive issues **Response SLA**: We aim to respond within 48 hours and issue fixes within 7 days for critical issues. ## Security Resources This guide maintains comprehensive security documentation: - **[Security Hardening Guide](./guide/security-hardening.md)** — MCP vetting, injection defense, audit workflows - **[Threat Database](./machine-readable/threat-db.yaml)** — 18 CVEs, 341 malicious skills - **[Security Hooks](./examples/hooks/)** — 30 production hooks (bash + PowerShell) - **[Security Commands](./examples/commands/)** — `/security-check`, `/security-audit`, `/update-threat-db` ## Security Maintenance **Threat Database Updates**: The threat intelligence database is updated based on: - CVE announcements and security advisories - Community reports of malicious skills/MCP servers - Anthropic security bulletins - Academic research (e.g., prompt injection papers) **Audit Schedule**: - Weekly review of new MCP servers and skills - Monthly audit of security documentation accuracy - Quarterly full threat database refresh **Last Updated**: 2026-02-11 (v3.26.0) ## Coordinated Disclosure If you're a security researcher and find issues affecting multiple repositories in the Claude Code ecosystem: 1. Email us first (coordinated disclosure preferred) 2. We'll coordinate with other maintainers if needed 3. Public disclosure timing: 90 days or after fix, whichever comes first ## Acknowledgments We thank security researchers who have contributed to improving this guide's security content through responsible disclosure. --- **Author**: [Florian BRUNIAUX](https://github.com/FlorianBruniaux) | Founding Engineer [@Méthode Aristote](https://methode-aristote.fr) **Guide License**: [CC BY-SA 4.0](./LICENSE)